Spin Torque and Magnetic Tunnel Junctions

Ed Myers, Frank Albert, Ilya Krivorotov, Sergey Kiselev, Nathan Emley, Patrick Braganca, Greg Fuchs, Andrei Garcia, Ozhan Ozatay, Eric Ryan, Jack Sankey, John Read, Phillip Mather, Dan Ralph Jordan Katine and Daniele Mauri (HGST)

Outline
Spin torque switching in spin valves
Switching speeds Asymmetry of switching currents (spin torque and spin accumulation) Reducing switching current levels

Non-uniform spin torque systems

Switching by concentrated spin current injection Vortex spin torque oscillator

Spin torque in magnetic tunnel junction

Probing spin torque as function of tunnel junction bias

Realizing Spin Transfer Effects

Cu free layer fixed layer Cu Py (2 nm) Cu (6 nm) Py (12 nm) Cu free layer fixed layer Cu Py (2 nm) AlOx (~0.7 nm Py (12 nm)

Nanopillar GMR

SPIN VALVE Low impedance ~ 0.01 -m2 GMR (R/R) ~ 10-20%

Nanopillar MAGNETIC TUNNEL JUNCTION

High impedance ~ 1 - 100 -m2 GMR (R/R) ~ < 50-90+% (varies with barrier thickness)

Critical current densities quite similar in good spin valves and MTJs High polarization of MTJs may give a ~ 2x advantage
Conventional ferromagnet spin transfer devices require lateral dimensions 250 nm to avoid significant self-field effects from required current levels Practical issues for spin-torque switching: speed, switching currents, impedance

Spin Transfer Driven Magnetic Reversal

Cu free layer fixed layer Cu

Py (2 nm) Cu (6 nm) Py (12 nm)

~40 nm

~120 nm T = 4.2 K

Nanopillar Spin-Valve
5.5
5.5

dV/dI [Ohm]

5.3

dV/dI [Ohm]

5.3

5.1 0 600 1200

5.1 -1 -0.5 0 0.5 1

Magnetic Field [G]

Current [mA]

Time Resolved Measurements of Nanomagnet Dynamics

Challenges
Time-resolved measurements In standard nanopillar devices, initial require devices with a non-zero direction of spin torque is determined by a angle between the free and the random thermal fluctuation from equilibrium. fixed layers. This leads to a random phase of the precessional dynamics. V(t) < 1 mV, t ~ 10-100 psec

st ~ m I m = m 2 I sin ( )
st
M m T>0 M
fixed layer

st
m
free layer

fixed layer

free layer

sin( ) << 1

sin( ) 1

Measurements of Spin-Transfer Dynamics

Step Generator

+25 dB

Sampling Oscilloscope

dc

I. N. Krivorotov et al. Science 307, 228 (2005).

~ 130 nm
Cu free layer fixed layer Py (4 nm) Cu (8 nm) Py (4 nm) IrMn (8 nm)

HEB

~ 60 nm

Cu

HEB = exchange bias field

Exchange biasing of the fixed Py layer at 45 to the easy axis results in a non-zero initial angle between magnetic moments of the fixed and free layers. This establishes a well-defined phase for precessional dynamics of the magnet.

Equilibrium Configuration of Magnetization

6.1

Sample 2
6.05

1 cos 2 ( / 2 ) R = R0 + R 1 + cos 2 ( / 2 )
0~ 35

dV/dI (Ohm)

= 0.5;

Heb = 1.5 kG

5.95

Mfixed

5.9 -400

0
-200 0 200 400 600

Filed (G)

Mfree Happlied

- data - Stoner-Wolfarth fit

High Speed Spin Torque Switching

ln 2 0 switching time1 = I Ico

0 ~ initial angle between magnetizations -set by thermal fluctuations or magnetic pinning Ic0 is T= 0 critical current

Spin polarized current must deliver sufficient spin angular momentum to nanomagnet to reverse magnetic moment. Hence (I Ic0)x = constant

Faster reversal requires larger Iswitch

1J.Sun,

Phys Rev B. 62, 570 (2000)

How fast is spin-transfer-driven switching?

Step Generator +25 dB

dc

Sampling Oscilloscope

Measure time dependent response of nanopillar resistance to step pulse.

Switching time < 1 ns at high pulse amplitude

I. N. Krivorotov et al. Science 307, 228 (2005).

Critical Current for Spin Torque Switching Ico+ = e Ms Vol [H + Han + 2 Ms ] / hg(0) 2 e Ms2 Vol / h g(0) Ico- = e Ms Vol [H - Han - 2 Ms ] / hg() 2 e Ms2 Vol / h g() Jco+ 2 e Ms2 t / h g(0); Han = shape anisotropy field
4 Ms out of plane demagnetization field Han top view

Jco+ 2 e Ms2 t / h g()

t = nanomagnet thickness, =Gilbert damping parameter, Ms = magnetization

To reduce Jco - reduce t, Ms and/or but must maintain nanomagnet stability This requires UK = MsHan Vol

/2 > 50 kBT - ten year bit stability

Decreasing Switching Currents

MRAM requirement: Bit lifetime ~ 10 years U0 = 1 eV at RT With heating to 100 C U0 = 1.3 eV

Ic Ms2 (Vol) U0 HanMs(Vol) Han ~ Ms(t/t0)

U0

~40 nm

Minimize Ms and sample volume Use shape anisotropy to maximize Hk thick and elongated 4.5 nm Py : U0,P-AP=0.85 eV, Ic0+ = .42 mA U0,AP-P=0.73 eV, Ic0- = .39 mA

~120 nm

Ic0 = zero-temp critical current. Need Ico < 100 A Need to decrease damping and improve micromagnetics

Spin torque switching currents of low Ms free layers

Pulse Generator +25 dB

Pulse-response measurements
1.0 0.8 0.6 0.4 0.2 0.0 0.0 100 ns 30 ns 10 ns 3 ns 1 ns

dc

Apply current pulse to device. Determine if pulse has switched device. Increase pulse duration until probability of switching goes to unity. Increase current pulse amplitude and repeat.

Switching Probability

0.5

1.0 1.5 2.0 Pulse Amplitude (mA)

2. 5

Comparison with Single Domain LLG Simulations

1.0

Switching Probability

1.4 1.2

0.8
100 ns 3 ns 1 ns
(mA)

0.6 0.4 0.2 0.0 0.5 1.0 1.5 2.0

1.0 0.8 0.6 0.4 0.2 0.0 0.2 0.4 0.6

simulations data

50%

p-ap ap-p
0.8
-1

1.0

-1

(ns

2.5

Pulse Amplitude (mA)

Fitting to LLG simulation yields empirical spin-torque function and damping N.B. Similar AP-P and P-AP switching currents in these devices Braganca et al. APL 05

Spin Transfer Torque Function

dm dm B g ( ) (m I m ) = (m H eff ) m + 2 dt dt e m sin( )
0.3 0.25 0.2

0.15 0.1 0.05 0

Ic, P - AP

g() Slonczewski 1996 g() Xiao, et al. g() Cornell (exp.)

See also: Manschot et al., APL.2004 Barnas et al. PRB 2005

/2 ~ g(0) ; Ic, AP - A ~ g()

A sin( ) g ( ) = 1 + B cos( )

effect of device geometry on g( )

spin accumulation affects?

Effect of Electrode Structure on Spin Torque

>> sf
0.3 0.25 0.2 0.15 0.1 0.05

> sf
Net electron flow

gold cap

Effect of Electrode Structure on Spin Torque

>> sf
0.3 0.25

> sf
Net electron flow

Fe-Mn cap

0.2 0.15 0.1 0.05

gold cap

FeMn

Pulsed Current Experiments

Pt Capped Devices
Standard Configuration 1.0 Switching Probability 0.8 0.6 0.4 0.2 0.0
1 ns pulse data 3 ns pulse data 10 ns pulse data 100 ns pulse data simulations A = 0.18 = 0.037

Inverted Configuration
A = 0.52 = 0.047

AP-P switching

1 ns pulse data 3 ns pulse data 10 ns pulse data 100 ns pulse data simulations

1 2 3 4 Pulse Amplitude (mA)

1 2 3 4 Pulse Amplitude (mA)

LLG simulations
Torque angular dependence

g ( ) =

A sin( ) 1 + B cos( )

A Torque amplitude from spin current and spin accumulation

Spin pumping enhancement in inverted samples Better spin sinking in extended Cu lead LLG fit deviation from data at large currents microwave oscillations

e-

1 B= 1+

I switch , AP P I switch , P AP

Au cap A
0.25-0.30

Fe-Mn cap
0.12-0.16

Pt cap
0.18-0.21

Pt inv.
0.45-0.52

0.02-0.19

0.32-0.33

0.11-0.23

0.08-0.13

0.025-0.030

0.033-0.037

0.033-0.037

0.047

A=0.18 B=0.23

A=0.52 B=0.13

Pt normal

Pt inverted

Spin-Transfer-Switching by Spatially Non-Uniform Currents

150x250nm 150x250 nmpillar pillar Goal: 30nm hole

A 3nm Al2O3 insulating barrier with a nano-orifice is inserted into a Cu/Py spin-valve nanopillar

Pt 30nm Cu Py 20nm Cu 8nm Result: SiO2 Al2O3 3nm

15-30nm aperture sizes SiO2

Py 5nm Cu

150nm

Spin-Transfer-Switching by Spatially Non-Uniform Currents

11.95 11.9 11.85 11.8
R( ) 12 11.95 11.9 11.85 11.8 11.75 11.7 11.65 -600 -400 -200 0 H(Oe)

Hc~37.5Oe R~253m T=4.2K

dV/dI()

11.75 11.7 11.65 -0.5 -0.4 -0.3

150m

The nano-aperture device requires much less current to induce switching than a nanopillar with uniform current flow. Current-induced switching may not result in full reversal of the nanomagnet

-0.2

-0.1

0.1

0.2

0.3

0.4

0.5

I(mA)

150 x 250 nm2 with 30 nm aperture 100 x 200 nm2 uniform current

P-AP Ic+ = 180 A P-AP Ic+ = 7.8 mA

AP-P Ic- = 50 AP-P Ic- = 4 mA

Jpillar ~ 4x105 A/cm2 Jhole~1.6x107 A/cm2

R = 12

J ~ 1.2x107 A/cm2

R=3

3D OOMMF Simulations
OOMMF is a public software developed by M.J.Donahue and D.G. Porter from NIST The effect of spin torque was modeled using LLG equation with the Slonczweski term for each cell. The simulations were performed taking into account the Oersted field created by electron flow through a wire.

6 10 6 4 10 6 2 10
6

5 10 0
-7

-8

-1 10

0
-8

-5 10

0 5 10
-8

-5 10 1 10 - 7

-8

t=1.13ns

t=1.6ns

t=2.06ns

t=2.3ns

t=2.5ns

t=3.3ns

t=3.96ns

t=5.9ns

0.5 mA

Spin Transfer with Magnetic Tunnel Junctions

Ok for spin transfer

Bad TMR, Pinholes

Good TMR, too high resistance to do spin transfer.

Challenge: Tunnel barriers with high TMR that can withstand the currents necessary for switching, particularly for fast switching

0.1

10

100 100010000 RA ( m2)

Pt 30 nm Cu 5 nm CoFeB 2 nm AlO x 7-8 CoFeB 8 nm Ta 10 nm Cu 80 nm

56 nm

147 nm

Early Demonstrations with AlOx

20 CoFeB 6.5 Al + Oxygen 80 CoFeB There is a small TMR measured with DC resistance at switching currents. Wear-out of barriers a concern due to high critical currents/voltages

CoFeB=Co88.2Fe9.8B2

Minor Loop T= 77K

H = 387 Oe

T = 77 K Switching currents
Huai et. al., APL 84, 3118 (2004) Fuchs et. al., APL 85, 1205 (2004)

Increasing spin torque in MTJs with three magnetic layers

5 nm CoFe 6 nm Cu 4 nm Py ~0.8 nm AlOx 8 nm CoFe 20 nm Ta

Anti-aligned fixed layers

Spins from each fixed layer are in the same direction more spin torque

Aligned fixed layers

Spins from each fixed layer are in opposite directions almost no spin torque

Spin Transfer Switching in 3-layer MTJs

Anti-aligned fixed layers
P AP/P AP/P AP T=77K

Aligned fixed layers

P P

AP

AP

Strong spin torque

Ohmic heating reduces Hc, minimal spin torque

Ic,o+ = 0.290.01 mA (shape and size Ic,o- = -0.280.01 mA not optimized) Jc,o/t = (2.9 0.4) x106 A/(cm2-nm), reduced by 40% compared to a Py free layer with one fixed layer: 5x106 A/(cm2-nm)
Note the similarity of Ics
G. D. Fuchs et al., Appl. Phys. Lett. 86, 152509 (2005).

Questions regarding spin torque in MTJs

-0.3V 0 0.3V

Why does TMR decrease with increasing bias? How does bias affect spintransfer torque? What is the nature of spin polarized transport in MgO based MTJs at finite bias?

Models that describe TMR(V) must also be consistent with spin torque, Nst/I(I) and I(V)

How to measure torque vs. current

A thermally stable free layer can only provide a measure of the spin-torque at the switching bias

A thermally unstable free layer can provide a measure of spin-torque continuously as a function of bias by applying H and I so as to have opposing effects

Sample structure
CoFe 1 nm/Py 1.8 nm MgO 0.8 nm CoFe 1.9 nm Ru 0.7 nm CoFe 2.2 nm PtMn 15.4 nm Bottom pinned SAF nearly cancels the dipole field and has a very large exchange field (~2 kOe) Devices are patterned with a 2:1 aspect ratio Have a range of thermal activation barriers

CoFe = Co86Fe14 Py = Ni91.5Fe8.5

100 nm
Katine and Mauri - HGST Lacour et al, APL 85, 4681, (2004)

Experimental approach
Lifetime in thermal activation regime

P / AP

E = o Exp a k BT

H H dip 1 H c,o

I ( I ) 1 m I c , o

Positions of equal mean lifetimes if the efficiency is constant with bias Magnetic Field (H)

(I)=Scaling factor to parameterize Nst/I

variation with I - Spin Transfer Efficiency

Positions of equal mean lifetimes if efficiency decreases with increasing bias

Increasing Current (I)

E. B. Myers, et al, PRL 89, 196801 (2002). Z. Li and S. Zhang, PRB 68, 024404 (2003). I. N. Krivorotov, et al, PRL 93 166603 (2004).

H(I) data - Linear Response

TMR decreases by over 40%

Hd

H(I) data - Linear Response

TMR decreases by over 40%

Break in data crystalline anisotropy effect

Spin Transfer Efficiency

Data are consistent with less than a 10% decrease in spin torque efficiency out to the switching bias point (~ 0.3 V)

Tunnel Conductance Through MgO

s-like pd-like No s-like channels!

s-like decays in the electrode

No s-like channels!

W. H. Butler, X. G. Zhang, T. C. Schulthess, PRB 63, 054416 (2001). J. Mathon and A. Umerski, PRB 63, 220403 (2001).

MgO DOS Data

Fe / 2nm MgO[eb] Fe / 20 MgO[eb] Fe / 2nm MgO[rf] Fe / 20 MgO[rf] CoFeB / 2nm MgO[rf] 375C 1 hr CoFeB / 20 MgO[rf] 375oC 1hr

DOS DOS(A.U.) (A.U.)

5.5 eV

2 eV

-6

-5

-4

-3

-2

-1

NegativeTip Tip Bias (V) (V) Negative Bias

STM tunneling spectroscopy evidence for O vacancy defects in MgO barrier layers

Tunnel Conductance through MgO

Magnetic state dependent effective mass (decay length):
W. H. Butler, X. G. Zhang, T. C. Schulthess, PRB 63, 054416 (2001). J. Mathon and A. Umerski, PRB 63, 220403 (2001).

Simmons model fit:

* =1.350.05 map

m* p =0.820.02
Elastic scattering by barrier defects reduces the TMR

AP

(I)~const implies that:

conductance for each spin channel varies with bias at a rate proportional to the zero bias DOS. electron scattering rate from defects is not strongly spin dependent!

Symmetry of Critical Currents

Polarization term

g ( ) =

P(V ) 2[1 + P(V ) 2 Cos ]

Asymmetry term is present to convert Slonczewskis critical voltage (Vc) into a critical current (Jc). A better approximation:

g ( ) =

P(V = 0) TMR (V ) 2 1 + Cos 2 + TMR (V )

Diao et al., APL 87, 232502, (2005) Polarization term is a constant function of V, consistent with our study

P2 calculated from TMR(V)

Conclusions ST in MTJs
Spin-transfer torque per unit current is independent of bias within 10% up to 0.35 V (good news for spin-torque driven MRAM) Measurement brings new information to help understand the relationship between bias and spin-polarized tunneling Results are inconsistent with: Free-electron, split-band tunneling models Magnon emission models that reduce polarization factors Results are consistent with calculations due to Butler et al and Mathon et al for transport through ultra-thin MgO tunnel barriers allowing for defects in non-ideal tunnel barriers.

Fuchs et al., cond-mat/0510786